I'm new to Scene and have been processing a simple 2 scan project as practice. I've got four checkerboard tagets, visible from both set-ups, for which I have surveyed co-ordinates. My problem is that Scene won't recognise the centre of the checkerboard targets properly. One scan is fine if I choose the checkboards manually using the picking tool. The other scan has a slightly poorer view of the targets, but by no means bad (the targets were put up by a Focus 3D distributor who was demoing the gear - he also chose the scan resolution. In the 'poorer' scan Scene will place the checkerboard 'object' off centre by 20-30mm, or wildy wrong - at the edge of the paper the target was printed on for example. I can't seem to overide this, or force Scene to only look within a selection box I define. Any help? If it is simply poor target/scanner geometry I'll be worried - it won't bode well for work I would like to use the scanner for. The range from scanner to targets is between 5m and 10m, 1/4 resolution.
All I can do at the moment I think is pick the scan point closest to the centre of the checkerboard in my 'poor' scan and 'force' place the scans with object naming. I will try and register the scan with corner points and lines, but really I'd like as much information as possible about the accuracy of the registration, related to my total station observations. I'm comparing the scan to a traditional total station elevation, hence my fussing.
Hi Pete. I'm fairly new to the Focus too. All I can think of is the angle of the targets to the scanner being too great. I try to avoid using checkerboards and only use spheres where possible. I'll put a few checkerboards out if i don't have enough spheres, but only use them as a fall back. As of yet, I've not had to use them. I did see thhe type of effect you're talking about, but that was when they were quite far away and not at a good angle. Maybe check the "matching" settings under "options".
Did you check the position of the found center points in 3D view? Sometime the QuickView and PlanarView are not showing the correct position because of the properties of these views and the perspective.
As this seems to be a re-occuring topic here, I would volunteer to have a look at the data and create some instructions for the forum here if possible. So if you have a way to provide the data for download (scan + survey points), just send me a private message.
Thank you very much for your help. This forum is fantastic, I'm very impressed. Re tips, I'd not seen that thread, thank you jon (I did search for people with similar problems...). I will try the tips mentioned in it shortly. A few things to add to initial explanation:
1: The angle of incidence from scanner to CB is about 45deg in one case, however in the other two it is near 90 2: Scene is definitely 'picking' the wrong point as CB centre, I have confirmed this in 3D view. Further to that when I pushed the registration through with 'wrong' CB objects, I was left with two CB objects for one target in the finished model, one object from each scan. I assume this is because the objects weren't close enough for Scene to realise they represented the same CB.
Now to each respondee in turn:
MarkL: AoI - In one case I suspect the angle may be too great, in others though I would be surprised (and a little worried) if this was the case. I'm keen to use CBs as much as possible in the field. I will be hiring a Focus 3D initially to supplement my other survey kit, so if I can co-ordinate a lot of CBs on a big site and then fly round with the scanner in one day I'll be happiest. It may be that I do need to use spheres as well though I suspect (particularily tripod mounted). I'll check 'matching' settings in the options, thanks for that tip.
3DForensics: Point no. on target - I don't know, yet. I will have to check shortly. When I chose scan points manually to try and approximate the centre of the CB I was able to get 'quite close' (not close enough mind). I would guess around 50 - it will vary between the failed targets as they their range varies too. It's also occured to me that I may have only loaded a portion of the total number of scan points, so I will check this - the green square was fully coloured which means all points I believe, but I had better check that.
jon: Thanks!
Oliver: As mentioned above I'm sure the points chosen by Scene are wrong. As for a trial process, thats very genorous of you, thanks. I'm sure many people will benefit. I should be able to get the scans + CB.csv on to DropBox - two .fls at 200MB each. I will PM you with a link once it is complete (we are in an old protected building here and our net connection is not fibre, so things like this take a while...)
I will report back here with my results. Thanks again everyone.
We are using the same checkerboards we were using when we were running 4.8, and we are finding issues with picking the centre of checkerboards.
Is this a 5.0 bug as this never happened with 4.8. As an example I can have a scan (1/4 res 4x quality) and i can be 10m away perpendicular to the target and it will simply refuse to select the centre regardless of where i pick around the target.
I have had a few scans where Scene wouldn't recognize the checkerboards correctly until I colorized the scan. It seems odd, but once I colorized the scan it picked them up fine. Might be worth a try.
We too have had this problem. Has anyone heard of any kind of resolution to this issue? It appears to be intermittent, and can work in one scan and not in the next. We are using 4.8 64 bit (have not upgraded to 5!).
chris_e wrote:We too have had this problem. Has anyone heard of any kind of resolution to this issue? It appears to be intermittent, and can work in one scan and not in the next. We are using 4.8 64 bit (have not upgraded to 5!).
Thank you,
Chris
I never had this problem in scene 4.8, only 5.0.
We did however run out of white cloth tape once and had to use silver which led to it picking the wrong part of the targets in 4.8.
